doi:10.1111/jpc.

14600

ORIGINAL ARTICLE

Development of a paediatric central venous access device

database: A retrospective cohort study of practice evolution and
risk factors for device failure
Tricia M Kleidon ,1,2,3 Claire M Rickard,2,3 Jessica A Schults,1,2,3 Gabor Mihala,2,4,5 Craig A McBride,2,3,6
John Rudkin,7 Brett Chaseling1 and Amanda J Ullman 2,3
Departments of 1Anaesthesia and Pain Management, 6Paediatric Surgery, and 7Information and Technology, Queensland Children’s Hospital, 2Alliance for
Vascular Access Teaching and Research (AVATAR) Group, 3School of Nursing and Midwifery, and 4Centre for Applied Health Economics, Menzies Health
Institute Queensland and 5School of Medicine, Griffith University, Brisbane, Queensland, Australia

Aim: To describe practice evolution, complications and risk factors for multiple insertion attempts and device failure in paediatric central venous
access devices (CVADs).
Methods: A paediatric retrospective cohort study using prospectively collected data from CVAD database 2012–2014. Data included were patient (i.e. age,
condition), insertion (i.e. indication, device, technique) and removal (complications, dwell). Descriptive statistics and incidence rates were calculated per
calendar year and compared. Risk factors for multiple insertion attempts and failure were explored with logistic regression and cox regression, respectively.
Results: A total of 1308 CVADs were observed over 273 467 catheter-days in 863 patients. Multiple insertion attempts remained static (14%) and sig-
nificantly associated with non-haematological oncology (odds ratio 2.19; 95% confidence interval (CI) 1.08–4.43), respiratory (3.71; 1.10–12.5), gastroen-
terology (4.18; 1.66–10.5) and other (difficult intravenous access) (2.74; 1.27–5.92). CVAD failure decreased from 35% (2012) to 25% (2014), incidence
rate from 1.50 (95% CI 1.25–1.80) to 1.28 (1.06–1.54) per 1000 catheter-days. Peripherally inserted CVAD failure was significantly associated with lower
body weight (per kilogram decrease, hazard ratio (HR) 1.02; 95% CI 1.00–1.03), cephalic vein (1.62; 1.05–2.62), difficult access (1.92; 1.02–3.73), sub-
optimal tip placement (1.69; 1.06–2.69) and gastroenterology diagnosis (2.27; 1.05–4.90). Centrally placed CVAD failure was significantly associated
with younger age (per year, HR 1.04; 95% CI 1.00–1.07), tunnelled device (3.38; 2.41–4.73) and gastroenterology diagnosis (1.70; 1.06–2.73).
Conclusions: While advancement in CVAD practices improved overall CVAD insertion and failure outcomes, further improvements and innova-
tion are necessary to ensure improved vessel health and preservation for children requiring CVAD.

Key words: central venous catheter; clinical registry; paediatrics; peripherally inserted central catheter; quality care; vascular access.

What is already known on this topic What this paper adds

1 One in four central venous access devices (CVADs) fail.
2 Vessel insufficiency threatens the survival of vascular-access-
dependent children.
3 CVAD complications and failure are often preventable with a
coordinated, multifactorial and interdisciplinary approach to
improve device insertion and care.
1 Interdisciplinary practice changes are challenging; however, this
study demonstrates that CVAD complications and failure are pre-
ventable, but require a multifactorial (surveillance, education,
uptake of new technology) and interdisciplinary (insertion, vascu-
lar access specialist, infectious diseases) approach to achieve this.
2 The shift to ultrasound-guided vascular access involves a challeng-
ing learning curve, which might initially be met with resistance
from clinicians previously expert at their preferred technique due
to the associated risk of failed insertion attempts and increased
complications associated with learning a new technique.
3 Analysis of a large paediatric vascular access data set for clini-
cians to benchmark outcomes and evaluate similar quality initia-
tives and improve practice.

Correspondence: Ms Tricia Kleidon, Department of Anaesthesia and Pain Management, Queensland Children’s Hospital, Queensland, Level 7, 501 Stanley
Street, South Brisbane, Qld. 4101. Fax: +61 73068 4419; email: tricia.kleidon@health.qld.gov.au
Conflicts of interest: TM Kleidon, CM Rickard and AJ Ullman: Griffith University has received unrestricted investigator-initiated research or educational grants
from product manufacturers (3M, Adhezion, Angiodynamics, Baxter, BBraun, BD-Bard, Centurion Medical Products, Cook Medical, Entrotech, Medtronic,
Smiths Medical). Griffith University has received consultancy payments from manufacturers (3M, BBraun, BD-Bard, ResQDevices, Smiths Medical). JA Schults,
G Mihala, CA McBride, J Rudkin and B Chaseling: No conflicts of interest to declare.
Accepted for publication 31 July 2019.

Journal of Paediatrics and Child Health (2019) 1

© 2019 Paediatrics and Child Health Division (The Royal Australasian College of Physicians)
Practice evolution: Paediatric vascular access
Approximately 25% of hospitalised children receive a central
venous access device (CVAD), for treatment administration.1
Insertion of CVADs can be challenging, and 25% of CVADs fail
prior to the completion of therapy,2 due to mechanical, infectious
and vascular complications.1,3 The prevalence of complications
such as central line-associated blood stream infection (CLABSI;
1.63 per 1000 catheter-days) and device occlusion or blockage
(e.g. peripherally inserted central catheters (PICCs); 2.2 per 1000
catheter-days) are especially concerning.2 The economic costs of
CVAD failure and complications are multifaceted, having both
direct (need for replacement, depletion of access sites, therapy
delays), and indirect health-care costs (staff and resource
utilisation). The Healthcare Cost and Utilization Project data esti-
mate a mean increase in costs of US$50 621 per device failure.4,5
The patient costs of CVAD failure include multiple and painful
reinsertion procedures, extended inpatient stays and long-term
vessel insufficiency.6
There is growing body of evidence on which we should base our
CVAD practices to reduce failure rates and associated patient harm.
This includes the incorporation of enhanced decontamination prod-
ucts (including chlorhexidine gluconate (CHG)),7 novel catheter
materials and designs,8 ultrasound guidance (USG),9 interventional
radiology techniques10 and alternative sites.10 Advances in medical
and surgical care have meant children are surviving previously fatal
illnesses, but with long-term health-care needs, consequently
maintaining vessel health into adulthood is increasingly important.11
Increased survivorship and a growing interest in vessel health and
preservation has driven an increased focus on interventions, which
reduce catheter failure and complications, such as catheter lock solu-
tions12 and catheter salvage,13,14 rather than replacement of catheters
affected by infection, occlusion and thrombosis.10,14 Despite this ren-
ewed focus, little is known about how well these CVAD practices
have been implemented into practice, and their impact on vessel
health and preservation in the acutely unwell child.
Internationally there are few data platforms to benchmark and
monitor CVAD practice. With the exception of bloodstream infec-
tion (BSI)15,16 few CVAD outcomes are routinely reported to exter-
nal organisations. This means clinically significant outcomes such as
occlusion are potentially under-recognised and lack benchmarking
across health services. Consequently clinicians, patients and organi-
sations lack data to transform patient care and optimise outcomes.16
In recognition of this gap, in 2012 we established a local CVAD
quality database within a tertiary referral paediatric hospital. Data-
base variables included patient, device and insertion variables, as
well as complication data, for all children requiring a CVAD. The
primary objective of this study was to describe changes in CVAD
insertion practices, and examine change in failure (incidence and
rate) over time. A secondary objective was to identify modifiable
and non-modifiable risk factors for multiple insertion attempts and
device failure. These data may support practice changes to mitigate
modifiable risk factors for CVAD failure.
Methods
Study design
A retrospective cohort study was undertaken, using data prospec-
tively entered into a paediatric hospital vascular access database
between calendar years 2012 and 2014.
2 Journal of Paediatrics and Child Health (2019)
© 2019 Paediatrics and Child Health Division (The Royal Australasian College of Physicians)
TM Kleidon et al.
Setting and population
The database was established in 2012 in a tertiary paediatric facility
(Royal Children’s Hospital, Brisbane, Australia), which provided a
full suite of paediatric inpatient and outpatient services (excluding
cardiothoracic surgery and neonatal intensive care) to children and
adolescents from birth to 18 years. Any child (0–18 years) who
attended the operating theatre suite for insertion of a CVAD
(i.e. PICCs, centrally inserted CVADs) was included in the database.
CVADs inserted outside the operating theatre (e.g. emergency
department, intensive care unit (ICU)) were excluded as the data-
base included only insertions in the operating suite. The study
received ethical approval from the Children’s Health, Queensland,
Human Research Ethics Committee (HREC/16/RCH/67).
Data collection and measures
From January 2012 to November 2014, CVAD-related data
were collected and entered into a standard data collection
form by the inserting clinician. Insertion data were collected
on insertion. Complication and removal data were collected
during routine clinical care by a vascular access specialist
(VAS). Clinical staff from subspecialty paediatric wards contrib-
uted data to the master database with data linked with Micro-
soft Excel spreadsheets (Microsoft Corporation, Washington,
DC, USA).
Database variables were developed based on outcomes and
quality measures previously reported in quality improvement
initiatives,17,18 and included patient demographic and clinical var-
iables, CVAD insertion, complication and removal data. Variables
were reviewed by multidisciplinary stakeholders, including VAS,
surgeons and infectious diseases, and were included in the final
data set if they were: (i) known to be associated with the out-
come and (ii) feasible to collect.16 Practices that contradicted cur-
rent evidence were identified by the VAS, and attempts to
influence practice change were made through a continuous cycle
of feedback. CVAD complication and failure outcomes were
defined in line with best practice and current evidence as per
Table S1 (Supporting Information).16,19 The confirmation of
venous thrombosis was made by an independent radiologist using
standard department protocols when a symptomatic patient was
referred for vascular imaging. Clinical staff obtained blood and
CVAD tip cultures on suspicion of infection, as per standard
hospital and pathology protocols.19 Diagnosis of CLABSI and
CVAD-related BSI was made by an independent infectious diseases
specialist, using the definition recommended at the time.20,21
Statistical analysis
Data were exported to Stata 14 (StataCorp; LLC, College Station,
TX, USA) for linkage and statistical analysis. Data cleaning was
performed in Excel and Stata. Lost-to-follow-up and censored
data (device removed for reasons other than failure) were mar-
ked in Stata syntax as required. Missing data were not imputed.
Participant demographic and clinical characteristics over
12-month time periods were presented using descriptive statistics.
The changes in clinical and device characteristics, over time were
tested with χ2 tests. Incidence rates for device failure were calcu-
lated, and Kaplan–Meier survival curves were generated. PICCs
TM Kleidon et al.
and other centrally placed CVADs were analysed separately to
allow for natural variation between the two procedures. Infre-
quent catheter insertions such as non-tunnelled CVC and
haemodialysis catheters were excluded from regression analyses.
Covariates (independent variables) were re-categorised and
dummy-coded as necessary. Covariates were selected and entered
in multivariable models at univariable P < 0.20 level. Multivari-
able models were generated manually, by stepwise removing
covariates at P ≥ 0.05 (backward method). The final model was
confirmed by re-entering the removed covariates one-by-one
(forward method). The proportional hazards assumption was
tested, and the Nelson-Aalen estimates were graphed by the Cox-
Snell residuals (graph not presented). The results of hypothesis
testing and regression analyses were compared with the results of
the same analyses on a reduced data set (first insertion per
patient). Results of these sensitivity analyses were reported only
if the polarity or statistical significance was found to be affected
by including multiple devices per patient in the main analysis.
Statistical significance was declared at P < 0.05 (two-sided).
Results
Patient and device characteristics
There were 1308 CVADs, in 863 patients, involving 273 467 catheter-
days over 35 months. Table 1 describes patient characteristics.
Temporal changes in CVAD insertion practices
There was no significant increase in the use of USG vessel punc-
ture over the study period; 53% in 2012 to 58% in 2014
(P = 0.133). The number of attempts required to successfully
insert CVADs did not change, 14–15% of patients continued to
require multiple insertion attempts. PICCs were the most com-
monly inserted device across the study period, increasing from
34 to 47% (P = 0.002), while tunnelled CVAD use decreased
from 32 to 20%. The use of single lumen catheters increased
from 60 to 72%, while the use of double lumen catheters
decreased from 33 to 21% (P < 0.001). Overall, 2% CHG in 70%
alcohol was the most commonly used skin preparation.
Risk factors associated with multiple insertion
attempts
As described in Table 2, the diagnostic groups; non-haematological
oncology (odds ratio (OR) 2.19; 95% confidence interval
(CI) 1.08–4.43), respiratory (OR 3.71; 95% CI 1.10–12.5), gastro-
enterology (OR 4.18; 95% CI 1.66–10.5) and other (e.g. difficult
intravenous access) (OR 2.74; 95% CI 1.27–5.92) were signifi-
cantly associated with an increased risk of multiple insertion
attempts in comparison to haematological malignancy for cen-
trally placed devices. For PICCs, use of the cephalic vein (OR 1.90;
95% CI 1.12–3.22) and other non-specified vein locations
(OR 2.6; 95% CI 1.38–5.24) were associated with a greater risk
for multiple insertion attempts. Additionally, dual lumen PICCs
had an increased risk for multiple insertion attempts compared
with single lumen devices (OR 0.21; 95% CI 0.06–0.68).
Journal of Paediatrics and Child Health (2019)
© 2019 Paediatrics and Child Health Division (The Royal Australasian College of Physicians)
Practice evolution: Paediatric vascular access
Temporal trends in CVAD complications and failure
Table 3 describes device outcomes by insertion period. Overall
27% of CVADs were removed due to device failure, which
decreased over time from 35 to 25% (P < 0.001; χ2 test). Occlu-
sion and suspicion of infection were consistently the most com-
mon causes of device failure, where reasons for failure were
recorded. Catheter removal due to suspected infection halved
over the study from 11 to 5% (P = 0.006; χ2 test). Catheters
requiring salvage due to infection also reduced from 12 to 5%
(P = 0.010; χ2 test).
Associations with PICC failure
As per Table 4, multivariable Cox regression identified several
patient-, provider- and device-related factors associated with
increased PICC failure. Lower body weight had a significantly
higher risk of PICC failure (per kilogram, hazard ratio (HR) 1.02;
95% CI 1.00–1.03; inverted for consistency). Non-modifiable risk
factors such as a gastroenterology diagnosis had a twofold
increase in failure (HR 2.27; 95% CI 1.05–4.90). Catheter tip out-
side the cavo-atrial junction (CAJ) was associated with increased
risk for PICC failure (HR 1.69; 95% CI 1.06–2.69). Venepuncture
of the cephalic vein, compared to basilic, was associated with
increased PICC failure (HR 1.62; 95% CI 1.05–2.62). Children
requiring PICC insertion due to difficult intravenous access, com-
pared to those without difficult access, had higher risk of PICC
failure (HR 1.92; 95% CI 1.02–3.73). The Kaplan–Meier curve
(Fig. 1a) demonstrates an increased rate of PICC failure in the lat-
ter years of the study.
Associations with centrally inserted catheter
failure
The association between device failure, and patient and device
insertion characteristics are described as per Table S2 (Supporting
Information). Tunnelled cuffed CVAD were more than three
times more likely to fail compared to totally implanted venous
port devices (TIVPDs) (HR 3.38; 95% CI 2.41–4.73; P < 0.001).
For every 1-year increase in age, a significant reduction in risk of
device failure was evident (HR 0.97; 95% CI 0.90–1.00). A gas-
troenterology diagnosis, when compared to haematology, non-
cancer, was positively associated with device failure (HR 1.70;
95% CI 1.06–2.73). The Kaplan–Meier curve (Fig. 1b) demon-
strates a reduction in failure over time.
Discussion
This study describes significant changes in CVAD insertion prac-
tices, and the associated CVAD insertion and failure over time in
a paediatric population at a single children’s hospital. Significant
changes in use of, skin antisepsis and judicious use of multi-
lumen catheters at insertion were seen. A reduction in CVAD
failure, from 35 to 24%, was evident (average 27%); however,
overall CVAD failure remains unacceptably high, and predomi-
nantly related to occlusion or suspected infection. Over time,
PICC failure increased which might be related to the increased
acuity of patients receiving PICC, as evidenced by the reduction
in centrally inserted CVADs. Reasons for failure were poorly
3
Practice evolution: Paediatric vascular access TM Kleidon et al.

Table 1 Patient and device characteristics†

n 2012, n (%) 2013, n (%) 2014, n (%)

Number of patients 863 209 317 337

Age group at last insertion, years 863
0–4 14 (7) 35 (11) 40 (12)
5–9 86 (41) 111 (35) 123 (36)
10–14 42 (20) 56 (18) 66 (20)
>15 14 (7) 21 (7) 24 (7)
Males 857 117 (56) 189 (60) 197 (59)
Weight at last insertion, kg, median (IQR) 804 19.0 (23.9) 20.0 (19.0) 18.0 (18.9)
Number of insertions 1308 349 494 465
Multiple insertion attempts 1286 49 (14) 68 (14) 67 (15)
Vein location 1284
Internal jugular 136 (39) 180 (37) 145 (32)
Basilic 76 (22) 118 (24) 118 (26)
Subclavian 76 (22) 84 (17) 72 (16)
Cephalic 22 (6) 43 (9) 51 (11)
Other 35 (10) 63 (13) 65 (14)
Insertion technique 1279
Ultrasound 183 (53) 283 (58) 261 (58)
Blind puncture 70 (20) 106 (22) 80 (18)
Surgical cut down 53 (15) 53 (11) 65 (15)
Rewire 24 (7) 26 (5) 24 (5)
Other 15 (4) 19 (4) 17 (4)
Catheter type and class 1290
Peripherally inserted central catheter 120 (34) 202 (41) 213 (47)
Totally implanted venous port device 85 (24) 122 (25) 120 (27)
Tunnelled CVC 112 (32) 109 (22) 90 (20)
Non-tunnelled CVC (Other) 11 (3) 38 (8) 17 (4)
Other 20 (6) 19 (4) 12 (3)
Insertion team 1306
Surgical 199 (57) 240 (49) 207 (45)
Anaesthetic 143 (41) 251 (51) 251 (54)
Other 6 (2) 3 (1) 6 (1)
Diagnosis 1290
Haematology (Malignancy) 76 (22) 109 (22) 129 (28)
Oncology 86 (25) 114 (23) 90 (19)
Respiratory 50 (15) 67 (14) 73 (16)
Infection 29 (9) 70 (14) 56 (12)
Gastroenterology 42 (12) 29 (6) 26 (6)
Other 55 (16) 100 (20) 89 (19)
Number of lumens 1279
One 205 (60) 322 (66) 326 (72)
Two 113 (33) 126 (26) 93 (21)
Three 22 (6) 37 (8) 31 (7)
Four 0 (0) 2 (0) 2 (<1)
Skin preparation 1279
2% CHG in 70% alcohol 148 (43) 252 (52) 230 (51)
10% Povidone iodine in aqueous 193 (57) 26 (5) 183 (41)
10% Povidone iodine in alcohol 0 (0) 209 (43) 27 (6)
Other 0 (0) 0 (0) 11 (2)
Tip placement 1282
Cavo-atrial junction 193 (56) 296 (61) 102 (23)
Superior vena cava 119 (35) 147 (30) 319 (71)
Other 32 (9) 44 (9) 30 (7)
Indication: Antibiotics 1280 98 (29) 165 (34) 141 (32)
Indication: Blood products 1280 9 (3) 22 (4) 0 (0)

(Continues)

4 Journal of Paediatrics and Child Health (2019)

© 2019 Paediatrics and Child Health Division (The Royal Australasian College of Physicians)
TM Kleidon et al. Practice evolution: Paediatric vascular access

Table 1 (Continued)

n 2012, n (%) 2013, n (%) 2014, n (%)

Indication: Chemotherapy 1280 162 (47) 224 (46) 207 (47)

Indication: Dialysis 1280 9 (3) 5 (1) 3 (1)
Indication: Difficult access 1280 10 (3) 51 (10) 25 (6)
Indication: Infusion 1280 15 (4) 34 (7) 41 (9)
Indication: Total parenteral nutrition 1280 54 (16) 50 (10) 38 (9)
Indication: Other(s) 1280 21 (6) 27 (5) 24 (5)

†Column frequencies (%) were calculated using the number of non-missing observations as denominator. CHG, chlorhexidine; CVC, central venous cathe-
ter; IQR, interquartile range.

recorded in the database (‘other’ being the default option), so
more comment cannot be made regarding reasons for failure.
The data set included multiple (up to 12) outcomes per partici-
pant, which did not appear to have influenced the findings of this
study in a meaningful way.
As evident in previous international literature,2 the risk of
CVAD complication was greatest for younger children with com-
plex, chronic pathology such as a gastroenterology diagnosis for
PICC (HR 2.27; 95% CI 1.05–4.90) and for centrally inserted cen-
tral catheter (HR 1.70; 95% CI 1.06–2.73), respectively.
Innovations to reduce complications and improve techniques of
catheter salvage in this vulnerable patient cohort, rather than
removal and replacement, are urgently required to allow preser-
vation of alternative access sites for future use.22 In 2014 the hos-
pital introduced taurolidine citrate for children with recurrent
CVAD BSI. Taurolidine citrate was indicated for patients with a
history of more than one CABSI, residing within the hospital and
did not have any medication infusion for at least 6 h. Longterm
sequelae for CVAD failure and complications for children with
chronic, vascular-access-dependent conditions are severe. In its

Table 2 Associations between multiple insertion attempts and patient/device insertion characteristics (logistic regression)

Central (n = 744), OR (95% CI) Peripheral (n = 531), OR (95% CI)

Univariable Multivariable Univariable Multivariable

Age (1 year increase) 0.95 (0.90–1.01)† * 0.98 (0.93–1.02) †

Female (Reference: Male) 0.96 (0.57–1.59) † 1.37 (0.90–2.08)† *
Vein location
Internal jugular Reference † NA †
Subclavian 0.95 (0.55–1.67) † NA †
Femoral 0.88 (0.20–3.87) † NA †
External jugular 1.79 (0.50–6.35) † NA †
Basilic NA † Reference Reference
Cephalic NA † 1.83 (1.09–3.09)‡ 1.90 (1.12–3.22)‡
Brachial NA † 1.61 (0.79–3.30)† 1.57 (0.76–3.23)
Other NA † 2.70 (1.40–5.23)‡ 2.69 (1.38–5.24)‡
Ins. technique (Reference: Ultrasound)
Blind puncture 1.03 (0.54–1.96) * 1.38 (0.63–3.06) †
Surgical cut down 1.89 (1.03–3.46)‡ * NA †
Ins. team: anaesthetics (Reference: Surgical) 0.68 (0.30–1.52) † 1.87 (0.23–15.4) †
Diagnosis (Reference: Haematological malignancy)
Oncology 2.19 (1.08–4.43)‡ 2.19 (1.08–4.43)‡ 0.55 (0.17–1.84) ‡
Respiratory 3.71 (1.10–12.5)‡ 3.71 (1.10–12.5)‡ 1.29 (0.67–2.48) *
Infection NA NA 0.59 (0.29–1.23)† *
Gastroenterology 4.18 (1.66–10.5)‡ 4.18 (1.66–10.5)‡ 0.70 (0.26–1.88) *
Other 2.74 (1.27–5.92)‡ 2.74 (1.27–5.92)‡ 0.61 (0.28–1.33) *
Number of lumens (Reference: One)
Two 0.99 (0.59–1.67) * 0.33 (0.13–0.85)‡ 0.21 (0.06–0.68)‡
Three 0.42 (0.15–1.21)† * NA NA

*Statistically significant at P < 0.20; **Statistically significant at P < 0.05. †Not eligible for multivariable analysis at P ≥ 0.20. ‡Excluded from the multivar-
iable model at P ≥ 0.05. CI, confidence interval; ins, insertion; NA, not applicable or cannot be calculated, OR, odds ratio; Reference, reference
category.

Journal of Paediatrics and Child Health (2019) 5

© 2019 Paediatrics and Child Health Division (The Royal Australasian College of Physicians)
Practice evolution: Paediatric vascular access TM Kleidon et al.

Table 3 Device outcomes

n 2012, n (%) 2013, n (%) 2014, n (%)

Number of removals 1308 349 494 465

Reason for removal† 1296
End of treatment 212 (63) 339 (69) 295 (63)
Occlusion (including fibrin sheath) 31 (9) 36 (7) 44 (9)
Still in situ 7 (2) 41 (8) 56 (12)
Suspected infection 37 (11) 41 (8) 22 (5)
Other 50 (15) 37 (7) 48 (10)
Failure (at removal): All groups†‡ 1296 118 (35) 114 (23) 114 (25)
Failure (at removal): Tunnelled/Implanted†‡ 626 87 (47) 63 (27) 51 (24)
Failure (at removal): PICC†‡ 535 19 (16) 41 (20) 54 (25)
Failure (at removal): Other types†‡ 135 12 (38) 10 (16) 9 (21)
Dwell time (Total), device-days 12 888 78 030 109 322 86 115
Dwell time, days, median (IQR): All groups 12 888 64 (14–273) 33 (10–278) 26 (10–261)
Dwell time, days, median (IQR): Tunnelled/Implanted 623 224 (98–467) 294 (131–819) 305 (116–674)
Dwell time, days, median (IQR): PICC 533 14 (9–26) 13 (8–22) 12 (7–17)
Dwell time, days, median (IQR): Other 132 7 (2–30) 5 (2–8) 7 (1–17)
IR of failure at removal§¶ 339 1.50 (1.25–1.80) 1.02 (0.85–1.23) 1.28 (1.06–1.54)
Devices with any linked (non-removal) complications† 1010 226 429 355
Devices with CLABSI† 1010 28 (12) 41 (10) 19 (5)
Devices with medically significant bacteria† 1010 14 (6) 17 (4) 5 (1)
Devices with breakage† 1010 16 (7) 16 (4) 4 (1)
Devices with occlusion† 1010 14 (6) 6 (2) 12 (3)
Devices with thrombosis† 1010 5 (2) 7 (2) 3 (1)
Devices with pulled out† 1010 4 (2) 8 (2) 0 (0)
Devices with other† 1010 4 (2) 8 (2) 6 (2)

†Frequencies and column percentages shown. ‡Coded as 1 for occlusion, suspected infection, dislodgement, breakage, thrombosis or other, and 0 for
end of treatment and censored events. §Per 1000 device-days. ¶Including 95% confidence interval. CLABSI, central line associated blood stream infec-
tion; IQR, interquartile range; IR, incidence rate; PICC, peripherally inserted central catheter.

most extreme scenario, loss of central venous catheter or absence
of an accessible central venous pathway can mean loss of life.23
Multifactorial influences are likely responsible for the 50%
reduction in catheter removal due to suspected infection, as until
2014, catheter removal to treat CLABSI was considered neces-
sary. Now the microorganism responsible for CLABSI influences
the decision to remove the catheter or attempt catheter salvage
with a prolonged course of empiric and direct antibiotic therapy,
supplemented by catheter lock.24,25 The Infectious Disease Society
of America guidelines recommend long-term catheters be
removed from patients with CLABSI associated with severe sep-
sis; suppurative thrombophlebitis; endocarditis; CLABSI that con-
tinues despite 72 h of targeted antimicrobial therapy; or
infections due to Staphylococcus aureus, Pseudomonas aeruginosa,
fungi or Mycobacteria spp.25 Catheter salvage is crucial for preser-
vation of long-term venous access in paediatric patients with
complex and chronic disease.26
For the past two decades, CLABSI prevention has received
increased attention with several quality improvement studies
demonstrating how simple and practical interventions markedly
reduce infection-related complications.27 A reduction in CLABSI
from 12 to 5% (P = 0.010; χ2 test) was observed from the start of
this study in 2012 to its conclusion in 2014. Current efforts to
maintain low CVAD-related BSI include surveillance and CLABSI
benchmark targets.21 Much of the 2014 CLABSI reduction may
be credited to the introduction of Taurolock (TauroPharm
GmbH,Waldbüttelbrunn, Germany) (taurolidine 1.34% with cit-
rate 4%), a catheter lock solution introduced in that year for
patients at high CLABSI risk.12
Insertion practices naturally evolved over the course of the
study, including the preferential use of 2% CHG in 70% alcohol
compared to the traditional 10% aqueous povidone iodine for
skin disinfectant. The Centers for Disease Control and Prevention
guidelines20 prompted the initial practice change, which were
actioned by the VASs and disseminated to the hospital depart-
ments through education sessions. The initial delay in practi-
tioner uptake may have been related to the quality of the
evidence supporting CHG as a superior skin antisepsis.28 How-
ever, a recent trial by Mimoz et al.7 demonstrated CHG in alcohol
significantly reduced the incidence of catheter-related infections
(0.28 vs. 1.77 per 1000 catheter-days), in comparison to
povidone iodine in alcohol, in adult ICUs. To increase
generalisability, further research comparing the use of antiseptic
solutions prior to CVAD insertions, in adults and children in non-
ICU settings is needed.
The use of real-time ultrasound to gain vessel access for CVAD
insertion did not increase significantly over the study, and there
was no reduction in the number of attempts to successful vein
cannulation. Internationally, ultrasound is considered gold stan-
dard technique to guide vessel puncture, compared to traditional

6 Journal of Paediatrics and Child Health (2019)

© 2019 Paediatrics and Child Health Division (The Royal Australasian College of Physicians)
TM Kleidon et al. Practice evolution: Paediatric vascular access

technique, on the basis it may expose patients to more risk dur-

Table 4 Associations between peripherally inserted central catheter ing the learning curve.33
failure and patient/device insertion characteristics (Cox regression) CVAD failure is often preventable, as evident by improved
Hazard ratio (95% CI) CVAD insertion34 and management35 practices that have resulted
in a reduction in the proportion of failed devices.9,35,36 We
Multivariable observed several modifiable risk factors for CVAD failure, including
Univariable (n = 480) catheter tip placement and vessel accessed. Like previous studies,
Age (1 year increase) 0.94 (0.91–0.98)† ‡ we found a significant increase in catheter failure when the cathe-
Female (Reference: Male) 0.75 (0.51–1.10)† § ter tip was positioned outside the CAJ. A study of 2574 PICCs
Weight (1 kg increase) 0.98 (0.97–0.99)† 0.98 (0.97–1.00)¶ found all non-central PICC tip locations including midline (inci-
Multiple insertion attempts 1.57 (1.00–2.47)† § dence rate ratios 4.59; 95% CI 3.69–5.69), midclavicular (2.15;
(Reference: No) 1.54–2.98) and other (3.26; 1.72–6.15), compared with central tip
Vein location (Reference: Basillic)† location, were associated with an increased risk of complications.37
Cephalic 1.40 (0.88–2.24) 1.62 (1.05–2.62)¶ In children, careful attention to catheter tip position is vital to
Brachial 0.68 (0.27–1.72) 0.79 (0.31–2.00) ensure the CVAD remains functional and minimise the number of
Other 1.69 (1.00–2.85) 1.39 (0.81–2.39) catheter replacements required during their lifetime.
Insertion: Blind puncture 1.10 (0.51–2.39) †† When the cephalic vein was accessed to insert a PICC, device
(Reference: US) failure increased significantly compared to placement via the
Diagnosis (Reference: Respiratory)† basilic vein. The preferential use of the cephalic vein is likely due
Infection 1.88 (0.93–3.81) 1.66 (0.79–3.48) to its superficial location on the lateral side of the upper arm, pro-
Haematology 1.73 (0.83–3.61) 1.71 (0.80–3.62) viding an easy target when ultrasound technology was not
Gastroenterology 2.24 (1.09–4.58) 2.27 (1.05–4.90)¶ used.38 However, we observed with a small increase in the
Oncology 2.37 (0.98–5.73) 2.11 (0.85–5.20)
uptake and mastery of USG venepuncture, a conservative
Other 2.94 (1.53–5.62) 1.88 (0.89–4.01)
increase in preferential puncture of the basilic vein.
Number of lumens: Two 0.81 (0.46–1.42) ††
Numerous non-modifiable risk factors were observed, which
(Reference: One)
significantly impacted CVAD failure, including age, weight and
Tip placement: Other 1.50 (0.98–2.31)† 1.69 (1.06–2.69)¶
diagnosis. These findings align with existing studies which dem-
(Reference: Cavo-atrial junction)
onstrate increased CVAD complications in certain diagnostic
Indication (Reference: No)
Antibiotics 0.68 (0.45–1.03)† § groups such as oncology and haematology, catheter types
Difficult access 2.15 (1.26–3.69)† 1.92 (1.02–3.73)¶ (PICCs) and subsequent catheters.1 Other non-modifiable risk
Infusion 2.04 (1.06–3.92)† § factors such as tunnelled cuffed CVADs, compared to TIVPD,
Total parenteral nutrition 1.41 (0.92–2.19)† § increase the risk of catheter failure. Insertion of TIVPD is not rec-
Other‡‡ 1.06 (0.68–1.66) § ommended or practical for children who require continuous infu-
sions such as parenteral nutrition or multiple infusions.39 This
†LR test statistically significant at P < 0.20. ‡Ineligible for multivariable demonstrates the gap in current strategies for the complex vascu-
analysis due to correlation with another covariate. §Dropped from mul- lar access needs of paediatric chronic disease sufferers. Although
tivariable model at Wald test P ≥ 0.05. ¶Statistically significant at Wald reducing vessel trauma by improving first attempt insertion suc-
test P < 0.05; ref = reference category. ††Ineligible for multivariable cess is an important focus of current vascular access studies,30 fur-
analysis at LR test P ≥ 0.20. ‡‡Includes apheresis, blood products, che-
ther interdisciplinary, co-ordinated approaches to CVAD insertion
motherapy, dialysis, inotropes and other. CI, confidence interval; LR,
and care are needed to implement and evaluate additional initia-
likelihood ratio; US, ultrasound.
tives for patients with lifelong vascular access needs to maximise
CVAD longevity and vessel patency.

landmark and surgical cut-down techniques.9,29 First attempt suc-
cess (OR 2.09; 95% CI 1.26–3.46; P ≤ 0.001), reduced procedural
complications (OR 0.47; 95% CI 0.24–0.91; P = 0.025)30 and zero
incidence of arterial puncture during CVAD insertion31 have been
reported when using USG, in comparison to blind puncture. Spe-
cific patient groups, such as centrally inserted CVADs with an
oncology, respiratory and gastroenterology diagnosis, were asso-
ciated with at least twofold greater odds of requiring multiple
insertion attempts. Reasons for the relatively low rate of USG are
not clear. This was an observational study, and the majority of
inserters were already on the plateau of their learning curve with
their preferred technique. This clinician learning curve can be
considerable and variable, as these clinicians are often highly
skilled in their previous insertion technique, and significant time
and practice is necessary for USG mastery.32,33 It may be that
these experienced clinicians are unwilling to learn a new
Implications for clinical practice
These findings have important implications for clinicians. As pre-
viously described, CVAD failure is high and this has significant
implications for the child, their family and the health service.
Interventions that might reduce the risk of device failure include:
use of ultrasound to insert catheters, ensuring catheter tip is
located at CAJ at insertion, use of 70% chlorhexidine and alcohol
skin antisepsis and taurolidine citrate catheter lock solution for
patients at risk of recurrent CABSI. Additionally, this study sub-
stantiates positive results of Corkum et al.26 and Zanwar et al.40
that advocate attempted catheter salvage in select patients rather
than immediate removal, as well as the use of catheter lock solu-
tion in high-risk patients.12 Additionally, although successful initi-
ation of practice change was slow, over time small improvement
in patient outcomes were observed and it is hoped that these

Journal of Paediatrics and Child Health (2019) 7

© 2019 Paediatrics and Child Health Division (The Royal Australasian College of Physicians)
Practice evolution: Paediatric vascular access
(a)
Fig. 1 (a) Kaplan–Meier curve of peripherally inserted central catheter failure. (
inserted device failure. ( ), 2012; ( ), 2013; ( ), 2014.
practice changes will continue, and so too will the trajectory of
improved patient outcomes. Furthermore, it is hoped that invest-
ment in USG training and education for clinicians will improve
clinically necessary skills. Large efficacy trials to confirm and fur-
ther validate evidence-based practice are necessary.
Strengths and limitations
This study has some limitations. The data came from a single ter-
tiary paediatric facility; thus, findings may not be generalisable.
CVADs inserted in intensive care or the emergency department
were not included, and all catheters were inserted in the operat-
ing suite. Catheter tip placement was confirmed using mobile
fluoroscopy at procedure, which might impact the accuracy of
catheter tip location. However, this study has many strengths
including its pragmatic design and data integrity. Vascular access
is often compartmentalised into inserters, and practitioners who
maintain them, making practice change difficult. Vascular access
experts collected the data, verifying the consistent use of
standardised definitions and the assessment of device outcomes
(e.g. CLABSI, thrombosis) by experts.
Conclusions
These findings add to the knowledge of risk factors for catheter
failure among paediatric patients. While CVAD insertion practice
is improving, and rates of CVAD failure are gradually declining,
device failure remains problematic and unacceptably high. A par-
adigm shift towards adoption of evidence-based insertion and
management practices to preserve the vessels of these vulnerable
patients is essential, to ensure children enter adulthood with an
intact vasculature. To further improve patient outcomes, CVAD
insertion, management, data collection and data sharing must
continue to evolve to maintain pace with new evidence.
8 Journal of Paediatrics and Child Health (2019)
© 2019 Paediatrics and Child Health Division (The Royal Australasian College of Physicians)
TM Kleidon et al.
(b)
), 2012; ( ), 2013; ( ), 2014. (b) Kaplan–Meier curve of centrally
References
1 Ullman AJ, Cooke M, Kleidon T, Rickard CM. Road map for improvement:
Point prevalence audit and survey of central venous access devices in
paediatric acute care. J. Paediatr. Child Health 2017; 53: 123–30.
2 Ullman AJ, Marsh N, Mihala G, Cooke M, Rickard CM. Complications
of central venous access devices: A systematic review. Pediatrics
2015; 136: e1331–44.
3 Chopra V, Anand S, Krein SL, Chenoweth C, Saint S. Bloodstream
infection, venous thrombosis, and peripherally inserted central cathe-
ters: Reappraising the evidence. Am. J. Med. 2012; 125: 733–41.
4 Harron K, Mok Q, Hughes D et al. Generalisability and cost-impact of
antibiotic-impregnated central venous catheters for reducing risk of
bloodstream infection in paediatric intensive care units in England.
PLoS One 2016; 11: e0151348.
5 Wilson MZ, Rafferty C, Deeter D, Comito MA, Hollenbeak CS. Attributable
costs of central line-associated bloodstream infections in a pediatric
hematology/oncology population. Am. J. Infect. Control 2014; 42: 1157–60.
6 Ullman AJ, Kleidon T, Cooke M, Rickard CM. Substantial harm associ-
ated with failure of chronic paediatric central venous access devices.
BMJ Case Rep. 2017; https://doi.org/10.1136/bcr-2016-218757.
7 Mimoz O, Lucet JC, Kerforne T et al. Skin antisepsis with
chlorhexidine-alcohol versus povidone iodine-alcohol, with and with-
out skin scrubbing, for prevention of intravascular-catheter-related
infection (CLEAN): An open-label, multicentre, randomised, controlled,
two-by-two factorial trial. Lancet 2015; 386: 2069–77.
8 Kleidon T, Ullman AJ, Zhang L et al. How does your PICCOMPARE? A
pilot randomized controlled trial comparing various PICC materials in
pediatrics. J. Hosp. Med. 2018; 13: 517–25.
9 Lau CS, Chamberlain RS. Ultrasound-guided central venous catheter
placement increases success rates in pediatric patients: A meta-anal-
ysis. Pediatr. Res. 2016; 80: 178–84.
10 Barnacle A, Arthurs OJ, Roebuck D, Hiorns MP. Malfunctioning central
venous catheters in children: A diagnostic approach. Pediatr. Radiol.
2008; 38: 363–78.
11 Hallam C, Weston V, Denton A et al. Development of the UK vessel
health and preservation (VHP) framework: A multi-organisational col-
laborative. J. Infect. Prev. 2016; 17: 65–72.
12 Clark J, Graham N, Kleidon T, Ullman A. Taurolidine-citrate line locks
prevent recurrent central line associated bloodstream infection in
pediatric patients. Pediatr. Infect. Dis. J. 2018; 38: e16–8.
TM Kleidon et al.
13 Sullivan PM, Merritt R, Pelayo JC, Ing FF. Recanalization of occluded
central veins in a parenteral nutrition-dependent child with no access.
Pediatrics 2018; 141 (Suppl. 5): S416–20.
14 Wolf J, Curtis N, Worth LJ, Flynn PM. Central line–associated blood-
stream infection in children: An update on treatment. Pediatr. Infect.
Dis. J. 2013; 32: 905–10.
15 O’Grady NP, Alexander M, Burns LA et al. Guidelines for the preven-
tion of intravascular catheter-related infections. Am. J. Infect. Control.
2011; 39: S1–34.
16 Schults JA, Rickard CM, Kleidon TM et al. Building a global, pediatric
vascular access registry: A scoping review of trial outcomes and qual-
ity indicators to inform evidence based practice. Worldviews Evid.
Based Nurs. 2018; 16: 51–59.
17 Fratino G, Molinari AC, Parodi S et al. Central venous catheter-
related complications in children with oncological/hematological
diseases: An observational study of 418 devices. Ann. Oncol. 2005;
16: 648–54.
18 Yang RY, Moineddin R, Filipescu D et al. Increased complexity and
complications associated with multiple peripherally inserted central
catheter insertions in children: The tip of the iceberg. J. Vasc. Interv.
Radiol. 2012; 23: 351–7.
19 Ullman AJ, Kleidon T, Gibson V et al. Central venous access device
SeCurement and dressing effectiveness (CASCADE) in paediatrics:
Protocol for pilot randomised controlled trials. BMJ Open 2016; 6:
e011197.
20 O’Grady NP, Alexander M, Burns LA et al. Guidelines for the preven-
tion of intravascular catheter-related infections. Clin. Infect. Dis. 2011;
52: e162–93.
21 Australian Commission on Safety and Quality in Health Care. Imple-
mentation Guide for Surveillance of Central Line Associated Blood-
stream Infection. Sydney: The Commission; 2015.
22 Baskin KM, Hunnicutt C, Beck ME, Cohen ED, Crowley JJ, Fitz CR.
Long-term central venous access in pediatric patients at high risk:
Conventional versus antibiotic-impregnated catheters. J. Vasc. Interv.
Radiol. 2014; 25: 411–8.
23 Baskin KM, Durack JC, Abu-Elmagd K et al. Chronic central venous
access: From research consensus panel to national multistakeholder
initiative. J. Vasc. Interv. Radiol. 2018; 29: 461–9.
24 Raad I, Hanna H, Maki D. Intravascular catheter-related infections:
Advances in diagnosis, prevention, and management. Lancet 2007;
7: 645–57.
25 Mermel AM, Bouza E, Craven DE, Flynn P, O’Grady N et al. Clinical
practice guidelines for the diagnosis and management of intravascu-
lar catheter-related infection: 2009 update by the Infectious Diseases
Society of America. Clin. Infect. Dis. 2009; 49: 1–45.
26 Corkum KS, Jones RE, Reuter CH, Kociolek LK, Morgan E, Lautz TB.
Central venous catheter salvage in children with Staphylococcus
aureus central line-associated bloodstream infection. Pediatr. Surg.
Int. 2017; 33: 1201–7.
27 Mermel LA. Prevention of intravascular catheter-related infections.
Ann. Intern. Med. 2000; 132: 391–402.
28 Chaiyakunapruk N, Veenstra DL, Lipsky BA, Saint S. Chlorhexidine
compared with povidone-iodine solution for vascular catheter-site
care: A meta-analysis. Ann. Intern. Med. 2002; 136: 792–801.
Journal of Paediatrics and Child Health (2019)
© 2019 Paediatrics and Child Health Division (The Royal Australasian College of Physicians)
Practice evolution: Paediatric vascular access
29 National Institute for Clinical Excellence. Guidance on the Use of
Ultrasound Locating Devices for Placing Central Venous Catheters.
London: The Institute; 2002.
30 Oulego-Erroz I, González-Cortes R, García-Soler P et al. Ultrasound-
guided or landmark techniques for central venous catheter place-
ment in critically ill children. Intensive Care Med. 2018; 44: 61–72.
31 Rivera-Tocancipa D, Diaz-Sanchez E, Montalvo-Arce CA. Ultrasound
versus anatomical landmarks: Immediate complications in the central
venous catheterization in children under 18 years of age. Rev. Esp.
Anestesiol. Reanim. 2018; 65: 366–372.
32 Grebenik CR, Boyce A, Sinclair ME, Evans RD, Mason DG, Martin B.
NICE guidelines for central venous catheterization in children. Is the
evidence base sufficient? Br. J. Anaesth. 2004; 92: 827–30.
33 Avanzini S, Guida E, Conte M et al. Shifting from open surgical cut
down to ultrasound-guided percutaneous central venous catheteriza-
tion in children: Learning curve and related complications. Pediatr.
Surg. Int. 2010; 26: 819–24.
34 Roebuck D, Kleidon T, CA. M, Barnacle A. Internal jugular vein (IJV)
patency after central venous (CV) access [abstract]. Pediatr. Radiol.
2005; 35 (Suppl. 1): s76.
35 Kleidon T, Illing A, Fogarty G, Edwards R, Tomlinson J, Ullman AJ.
Improving the central venous access devices maintenance process to
reduce associated infections in paediatrics: Evaluation of a practical,
multi-faceted quality-improvement initiative. Healthcare Infect. 2015;
20: 46–53.
36 Ista E, van der Hoven B, Kornelisse RF et al. Effectiveness of insertion
and maintenance bundles to prevent central-line-associated blood-
stream infections in critically ill patients of all ages: A systematic
review and meta-analysis. Lancet Infect. Dis. 2016; 16: 724–34.
37 Jumani K, Advani S, Reich NG, Gosey L, Milstone AM. Risk factors for
peripherally inserted central venous catheter complications in chil-
dren. JAMA Pediatr. 2013; 167: 429–35.
38 Bajzer CT. Venous drainage of the upper extremities. In: Bhatt DL,
ed. Guide to Peripheral and Cerebrovascular Intervention. London:
Remedica; 2004.
39 Pittiruti M, Hamilton H, Biffi R, MacFie J, Pertkiewicz M. ESPEN guide-
lines on parenteral nutrition: Central venous catheters (access, care,
diagnosis and therapy of complications). Clin. Nutr. 2009; 28: 365–77.
40 Zanwar S, Jain P, Gokarn A et al. Antibiotic lock therapy for salvage of tun-
neled central venous catheters with catheter colonization and catheter-
related bloodstream infection. Transpl. Infect. Dis. 2018; 21: e13017.
Supporting Information
Additional Supporting Information may be found in the online
version of this article at the publisher’s web-site:
Table S1. Definitions of central venous access device
complications.
Table S2. Associations between centrally inserted device failure
and patient/device insertion characteristics (Cox regression).
9